Abstract
In the EU project SHAREWORK, methods are developed that allow humans and robots to collaborate in an industrial environment. One of the major contributions is a framework for task planning coupled with automated item detection and localization. In this work, we present the methods used for detecting and classifying items on the shop floor. Important in the context of SHAREWORK is the user-friendliness of the methodology. Thus, we renounce heavy-learning-based methods in favor of unsupervised segmentation coupled with lenient machine learning methods for classification. Our algorithm is a combination of established methods adjusted for fast and reliable item detection at high ranges of up to eight meters. In this work, we present the full pipeline from calibration, over segmentation to item classification in the industrial context. The pipeline is validated on a shop floor of 40 sqm and with up to nine different items and assemblies, reaching a mean accuracy of 84% at Hz.
Highlights
SubtractionMerging and Voxelization SegmentationClassification x x x ** Point CloudSynchronization no x* yes x yes x* x no no*: only utility; **: projected image.Without synchronization, it is not guaranteed that corresponding data are processed, which may result in misalignment or malformed projections
We choose to provide the neural network with a voxel-filtered point cloud, as this is data already available in our pipeline
As the accuracy of the classification is directly dependent on the quality of the unsupervised segmentation, it is reasonable to test both modules separately on the same data set
Summary
It is not guaranteed that corresponding data are processed, which may result in misalignment or malformed projections. In the approximate time scheme, all arriving data are held back in queues. Once all queues have data, the last arrival is named the pivotal point. From this data set, time deltas are calculated toward sets in the other queues. Afterward, incoming images are filtered to get a foreground mask. The output of background subtraction is a binary mask, where foreground objects are denoted 1. The method is based on an adaptive mixture of Gaussians (MOG) model established by Zivkovic [17]. In contrast to the original method, we prevent the background from updating after converging to the original workspace
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
